fix: allow server not initialized error to be retried (#18300)

Since relaxing quorum the error across pools
for ListBuckets(), GetBucketInfo() we hit a
situation where loading IAM could potentially
return an error for second pool that server
is not initialized.

We need to handle this, let the pool come online
and retry transparently - this PR fixes that.
This commit is contained in:
Harshavardhana
2023-10-23 12:30:20 -07:00
committed by GitHub
parent bbfea29c2b
commit fd37418da2
3 changed files with 29 additions and 20 deletions

View File

@@ -372,6 +372,12 @@ func initAllSubsystems(ctx context.Context) {
}
func configRetriableErrors(err error) bool {
if err == nil {
return false
}
notInitialized := err.Error() == "Server not initialized, please try again"
// Initializing sub-systems needs a retry mechanism for
// the following reasons:
// - Read quorum is lost just after the initialization
@@ -392,7 +398,8 @@ func configRetriableErrors(err error) bool {
errors.As(err, &wquorum) ||
isErrObjectNotFound(err) ||
isErrBucketNotFound(err) ||
errors.Is(err, os.ErrDeadlineExceeded)
errors.Is(err, os.ErrDeadlineExceeded) ||
notInitialized
}
func bootstrapTraceMsg(msg string) {
@@ -813,10 +820,12 @@ func serverMain(ctx *cli.Context) {
}()
go func() {
r := rand.New(rand.NewSource(time.Now().UnixNano()))
if !globalDisableFreezeOnBoot {
defer bootstrapTrace("unfreezeServices", unfreezeServices)
t := time.AfterFunc(5*time.Minute, func() {
logger.Info(color.Yellow("WARNING: Taking more time to initialize the config subsystem. Please set '_MINIO_DISABLE_API_FREEZE_ON_BOOT=true' to not freeze the APIs"))
logger.Info(color.Yellow("WARNING: Initializing the config subsystem is taking longer than 5 minutes. Please set '_MINIO_DISABLE_API_FREEZE_ON_BOOT=true' to not freeze the APIs"))
})
defer t.Stop()
}
@@ -864,9 +873,18 @@ func serverMain(ctx *cli.Context) {
var buckets []BucketInfo
// List buckets to initialize bucket metadata sub-sys.
bootstrapTrace("listBuckets", func() {
buckets, err = newObject.ListBuckets(GlobalContext, BucketOptions{})
if err != nil {
logger.LogIf(GlobalContext, fmt.Errorf("Unable to list buckets to initialize bucket metadata sub-system: %w", err))
for {
buckets, err = newObject.ListBuckets(GlobalContext, BucketOptions{})
if err != nil {
if configRetriableErrors(err) {
logger.Info("Waiting for list buckets to succeed to initialize buckets.. possible cause (%v)", err)
time.Sleep(time.Duration(r.Float64() * float64(time.Second)))
continue
}
logger.LogIf(GlobalContext, fmt.Errorf("Unable to list buckets to initialize bucket metadata sub-system: %w", err))
}
break
}
})